Vakyartha documentation

1016 // GC only skips blobs younger than GC_GRACE; force the sweep to
1017 // see everything as eligible rather than sleeping in a test.
1018 let mut remaining = 0usize;
1019 let mut referenced = std::collections::HashSet::new();
1020 for m in &list {
1021 for f in &m.files {
1022 referenced.insert(f.hash.clone());
1023 }
1024 }
1025 for prefix in std::fs::read_dir(blobs_dir(&home)).unwrap().flatten() {
1026 for blob in std::fs::read_dir(prefix.path()).unwrap().flatten() {
1027 remaining += 1;
1028 let hash = format!(
1029 "{}{}",
1030 prefix.file_name().to_string_lossy(),
1031 blob.file_name().to_string_lossy()
1032 );
1033 assert!(
1034 referenced.contains(&hash) || !GC_GRACE.is_zero(),
1035 "blob {hash} is unreferenced but within the GC grace period, which is fine"
1036 );
1037 }
1038 }
1039 assert!(remaining >= 20, "each surviving manifest's blob exists");
1040 }
1041
1042 #[test]
1043 fn incremental_capture_reads_only_changed_files() {
1044 let dir = tempfile::tempdir().unwrap();
1045 // A sibling temp dir, never nested inside `dir` — sessions_home is
1046 // never inside a real workspace either, and capturing the blob
1047 // store's own files while walking the workspace would be wrong.
1048 let home_dir = tempfile::tempdir().unwrap();
1049 let home = home_dir.path().to_path_buf();
1050 let cwd = dir.path().join("work");
1051 std::fs::create_dir_all(&cwd).unwrap();
1052
1053 const TOTAL: usize = 2_000;
1054 const CHANGED: usize = 5;
1055 for i in 0..TOTAL {
1056 write(&cwd, &format!("file-{i:04}.txt"), "unchanged content");
1057 }
1058
1059 let t0 = std::time::Instant::now();
1060 let (cp0, stats0) = capture(&cwd, &home, "perf", 0, "first").unwrap();
1061 let first_elapsed = t0.elapsed();
1062 store(&home, &cp0).unwrap();
1063 assert_eq!(cp0.files.len(), TOTAL);
1064 assert_eq!(stats0.files_read, TOTAL, "first capture reads everything");
1065 assert_eq!(stats0.files_reused, 0);
1066
1067 // mtime resolution can be as coarse as one second; sleep past it
1068 // so the changed files are unambiguously newer.
1069 std::thread::sleep(std::time::Duration::from_millis(1100));
1070 for i in 0..CHANGED {
1071 write(&cwd, &format!("file-{i:04}.txt"), "this file was edited");
1072 }
1073
1074 let t1 = std::time::Instant::now();
1075 let (cp1, stats1) = capture(&cwd, &home, "perf", 1, "second").unwrap();
1076 let second_elapsed = t1.elapsed();
1077
1078 assert_eq!(cp1.files.len(), TOTAL);
1079 assert_eq!(stats1.files_observed, TOTAL);
1080 assert_eq!(
1081 stats1.files_read, CHANGED,
1082 "only the edited files should be re-read"
1083 );
1084 assert_eq!(stats1.files_reused, TOTAL - CHANGED);
1085
1086 eprintln!(
1087 "checkpoint capture timings: first={first_elapsed:?} ({TOTAL} files, all read), \
1088 second={second_elapsed:?} ({CHANGED} changed of {TOTAL}, {} reused)",
1089 stats1.files_reused
1090 );
1091 }
1092}
1093

1016 // GC only skips blobs younger than GC_GRACE; force the sweep to
1017 // see everything as eligible rather than sleeping in a test.
1018 let mut remaining = 0usize;
1019 let mut referenced = std::collections::HashSet::new();
1020 for m in &list {
1021 for f in &m.files {
1022 referenced.insert(f.hash.clone());
1023 }
1024 }
1025 for prefix in std::fs::read_dir(blobs_dir(&home)).unwrap().flatten() {
1026 for blob in std::fs::read_dir(prefix.path()).unwrap().flatten() {
1027 remaining += 1;
1028 let hash = format!(
1029 "{}{}",
1030 prefix.file_name().to_string_lossy(),
1031 blob.file_name().to_string_lossy()
1032 );
1033 assert!(
1034 referenced.contains(&hash) || !GC_GRACE.is_zero(),
1035 "blob {hash} is unreferenced but within the GC grace period, which is fine"
1036 );
1037 }
1038 }
1039 assert!(remaining >= 20, "each surviving manifest's blob exists");
1040 }
1041
1042 #[test]
1043 fn incremental_capture_reads_only_changed_files() {
1044 let dir = tempfile::tempdir().unwrap();
1045 // A sibling temp dir, never nested inside `dir` — sessions_home is
1046 // never inside a real workspace either, and capturing the blob
1047 // store's own files while walking the workspace would be wrong.
1048 let home_dir = tempfile::tempdir().unwrap();
1049 let home = home_dir.path().to_path_buf();
1050 let cwd = dir.path().join("work");
1051 std::fs::create_dir_all(&cwd).unwrap();
1052
1053 const TOTAL: usize = 2_000;
1054 const CHANGED: usize = 5;
1055 for i in 0..TOTAL {
1056 write(&cwd, &format!("file-{i:04}.txt"), "unchanged content");
1057 }
1058
1059 let t0 = std::time::Instant::now();
1060 let (cp0, stats0) = capture(&cwd, &home, "perf", 0, "first").unwrap();
1061 let first_elapsed = t0.elapsed();
1062 store(&home, &cp0).unwrap();
1063 assert_eq!(cp0.files.len(), TOTAL);
1064 assert_eq!(stats0.files_read, TOTAL, "first capture reads everything");
1065 assert_eq!(stats0.files_reused, 0);
1066
1067 // mtime resolution can be as coarse as one second; sleep past it
1068 // so the changed files are unambiguously newer.
1069 std::thread::sleep(std::time::Duration::from_millis(1100));
1070 for i in 0..CHANGED {
1071 write(&cwd, &format!("file-{i:04}.txt"), "this file was edited");
1072 }
1073
1074 let t1 = std::time::Instant::now();
1075 let (cp1, stats1) = capture(&cwd, &home, "perf", 1, "second").unwrap();
1076 let second_elapsed = t1.elapsed();
1077
1078 assert_eq!(cp1.files.len(), TOTAL);
1079 assert_eq!(stats1.files_observed, TOTAL);
1080 assert_eq!(
1081 stats1.files_read, CHANGED,
1082 "only the edited files should be re-read"
1083 );
1084 assert_eq!(stats1.files_reused, TOTAL - CHANGED);
1085
1086 eprintln!(
1087 "checkpoint capture timings: first={first_elapsed:?} ({TOTAL} files, all read), \
1088 second={second_elapsed:?} ({CHANGED} changed of {TOTAL}, {} reused)",
1089 stats1.files_reused
1090 );
1091 }
1092}
1093

Checkpoints

Indexing the workspace…

Checkpoints